1. Field of the Invention
The present invention relates generally to computer software. More particularly, the present invention relates to network management software.
2. Description of the Relevant Art
The field of network management involves the management of networked devices, often remotely. A computer network is a linked group of two or more computers. Generally, networks may be classified as Local-Area Networks (LANs) or Wide-Area Networks (WANs). In a LAN, the computers or devices are typically connected together within a “local” area such as a home, office, or group of offices. In a WAN, the computers or devices are typically separated by a greater distance and are often connected via telephone/communication lines, radio waves, or other suitable means of connection.
Networks are usually classified using three properties: topology, protocol, and architecture. Topology specifies the geometric arrangement of the network. Common topologies are bus, ring, and star configurations. A network's protocol specifies a common set of rules and/or signals, such as Ethernet or Token Ring, which the networked devices use to communicate with each other. A network's architecture typically specifies one of the two major types of network architecture: peer-to-peer or client/server. In a peer-to-peer networking configuration, there is no server, and computers simply connect with each other in a workgroup to share files, printers, services, and Internet access. Client/server networks often include a domain controller to which all of the networked computers log on. This server may provide various services such as centrally routed Internet access, e-mail, file sharing, printer access, and security services.
Many types of devices may be managed over a network, such as printers, scanners, phone systems, copiers, and many other devices and appliances configured for network operation. Typically, such devices are managed via requests and events. A request is a message sent to a managed object. A request may be sent by a manager application to a managed object to query the object about a particular parameter associated with the object. A request may also be sent to a managed object to modify a parameter of the object. Alternately, an event is a message originating with a managed object. Events may be sent by managed objects to signal some change of state of the managed object, or to communicate information about the managed object. Managing such devices tends to require that the data types of each device's control parameters and signals be well defined. For example, a networked printer might have a Boolean status parameter which indicates whether the device is currently on or off and a control parameter which turns the printer on or off. A manager application may send a request to determine the on/off status of the printer. Then, once the status is determined, say, to be off, a subsequent request may be sent to modify the control parameter to turn the printer on. The printer may also be capable of generating an alert signal indicating, for example, that the toner level is low. In this case, an event communicating that fact may be sent by the managed object (the printer) to the appropriate manager application.
The network management software should be able to read and write these data correctly in order to manage the device. To do this, information about the data is required. Such information is referred to as metadata, or “data about data.” Metadata may typically describe what type of data (string, integer, Boolean, structure) an object has and how the data are formatted. Metadata is essential for understanding information related to managed devices, as well as information stored in data warehouses. Typically, network management software manages a given device by storing and manipulating a representation of its pertinent data as a software object, herein referred to as a “managed object.” This object is the virtual representation of the device on the network.
FIG. 1a illustrates an example of typical elements of a telecommunications network. The telecommunications world is characterized by devices such as cell phones, cell phone towers and other kinds of towers 156, phone systems 151, faxes 152, routers 153, switches 154, satellite dishes 155, etc., which may be interconnected via networks 108a. The management of such a large number of devices results in a huge volume of event/request traffic, which must itself be managed. In response to the network management needs of this technology sector, a conceptual framework for telecom network management called TeleManagement Network (TMN) was developed by the Telecom Management Forum (TMF). TMN defines the relationship between basic network building blocks, such as network elements, different network protocols, and operations systems, in terms of standard interfaces. Generally, a TMN system includes Agent hardware 150, Manager software 170, and Agent software 160. The Agent hardware 150 includes the managed devices such as those shown in FIG. 1a. The Manager software 170 includes any application used to manage a networked device. These manager applications, or client applications, may be installed and executed on one or more client computer systems 171a, 171b, . . . , 171n. The Agent software 160 includes the software interface between the Manager software 170 (for communications via network 108b) and the Agent hardware 150 (for communications via network 108a). The Agent software 160 may be installed and executed on one or more server computer systems 161a, 161b, . . . , 161n. In some instances, the Agent software 160 and Manager software 170 may be installed and executed on the same computer system. The Agent software 160 may also reside, in whole or part, on the Agent hardware 150 itself.
One TMN approach to managing objects over a network is the Simple Network Management Protocol (SNMP), a set of protocols for managing complex networks. SNMP works by sending messages, called protocol data units (PDUs), to different parts of a network. SNMP-compliant devices, called agents, store data about themselves in Management Information Bases (MIBs) and return this data to the SNMP requesters. The metadata used by SNMP to describe managed object data variables includes the variable title, the data type of the variable (e.g. integer, string), whether the variable is read-only or read-write, and the value of the variable. SNMP works over the TCP/IP (Transport Control Protocol/Internet Protocol) communication stack. SNMP also uses UDP over IP, and also may support TCP over IP. It is widely held, however, that SNMP was developed as a simple “quick fix” and was never intended to be a permanent solution to network management. Consequently, one problem with SNMP is that the information it specifies is neither detailed nor well-organized enough to adequately serve the expanding needs of modern networking.
Another example of a TMN network management protocol is the Common Management Information Protocol (CMIP). In the U.S. the CMIP protocol is primarily run over TCP/IP, while in Europe it is generally run over the OSI (Open Systems Interconnection) communication stack and was designed to replace SNMP and address SNMP's shortcomings by providing a larger, more detailed network manager. Its basic design is similar to SNMP: Management requests, management responses, and notifications are employed to monitor a network. These correspond to SNMP's PDUs. CMIP, however, contains eleven types of messages, compared to SNMP's five types of PDUs.
In CMIP, variables are seen as complex and sophisticated data structures with many attributes. These include: variable attributes, which represent the variable's characteristics (e.g., its data type, whether it is writable); variable behaviors, or the actions of that variable that can be triggered; and notifications, or event reports generated by the variable whenever a specified event occurs (e.g., a terminal shutdown would cause a variable notification event).
As a comparison, SNMP only employs variable attributes and notifications, but not variable behaviors. One of the strongest features of the CMIP protocol is that its variables not only relay information to and from the terminal (as in SNMP), but they can also be used to perform tasks that would be impossible under SNMP. For instance, if a terminal on a network cannot reach its fileserver for a predetermined number of tries, then CMIP can notify the appropriate personnel of the event. With SNMP, a user would need to explicitly keep track of the number of unsuccessful attempts to reach the fileserver. CMIP thus results in a more efficient network management system, as less work is required by a user to keep updated on the status of the network.
A significant disadvantage of the CMIP protocol is that it requires more system resources than SNMP, often by a factor of ten. Thus, any move to CMIP from SNMP requires a dramatic upgrade in network resources. Another disadvantage with CMIP is that it is very difficult to program; the variable metadata includes so many different components that few programmers are generally able to use the variables to their full potential.
Both of the above protocols have been implemented in a number of programming languages, such as C, C++, and Java™. However, network management software which takes advantage of SNMP or CMIP must be written specifically for the language of the protocol implementation. In other words, SNMP-based and CMIP-based network management software is dependent upon a particular programming language and protocol implementation.
GDMO (Guidelines for Definition of Managed Objects) is a standard for defining objects in a network in a consistent way. With a consistent “language” for describing such objects as workstations, LAN servers, and switches, programs can be written to control or sense the status of network elements throughout a network. GDMO prescribes how a network product manufacturer must describe the product formally so that others can write programs that recognize and deal with the product. Using GDMO with ASN1, descriptions may be made of the class or classes of the object, how the object behaves, its attributes, and classes that it may inherit.
GDMO is part of the CMIP and also the guideline for defining network objects under TMN. The object definitions created using GDMO and related tools form a Management Information Base (MIB). GDMO uses Abstract Syntax Notation One (ASN1) as the rules for syntax and attribute encoding when defining the objects. Abstract Syntax Notation One is a language that defines the way data is sent across dissimilar communication systems. ASN1 ensures that the data received is the same as the data transmitted by providing a common syntax for specifying application layer (e.g., program-to-program communications) protocols. Each communications system contains a similar ASN1 encoding/decoding scheme written in the language used on that particular system. When one system wants to send data to another, the first system encodes the data into ASN1, sends the data, and the second system receives and decodes the data using the decoder written in the language used on that system.
In response to the difficulties presented by SNMP and CMIP, the Object Management Group (OMG) and Joint Inter-Domain Management (JIDM) have defined Interface Definition Language (IDL) for network management, which is used to access object instance data and may be used across a plurality of programming languages and across a plurality of platforms. JIDM IDL allows programmers to write only one set of interfaces for a particular object across multiple programming languages, rather than having to write a new set of interfaces for each programming language.
A middleware standard used extensively in network management is the Common Object Request Broker Architecture (CORBA), which is provided by the Object Management Group (OMG). CORBA specifies a system that provides interoperability between objects in a heterogeneous, distributed environment and in a way transparent to the programmer. Its design is based on the OMG Object Model, which defines common object semantics for specifying the externally visible characteristics of objects in a standard and implementation-independent way. In this model, clients request services from objects (which will also be called servers) through a well-defined interface. This interface is specified in the OMG Interface Definition Language (IDL).
In CORBA, a client accesses an object by issuing a request to the object. The request is an event, and it carries information including an operation, the object reference of the service provider, and actual parameters, if any. The object reference is an object name that reliably defines an object.
A central component of CORBA is the Object Request Broker (ORB). The ORB encompasses the communication infrastructure necessary to identify and locate objects, handle connection management, and deliver data. In general, the ORB is not required to be a single component; it is simply defined by its interfaces. The basic functionality provided by the ORB includes passing the requests from clients to the object implementations on which they are invoked. The ORB acts as the middleware between clients and servers. In the CORBA model, a client can request a service without knowing anything about what servers are attached to the network. The various ORBs receive the requests, forward them to the appropriate servers, and then hand the results back to the client.
In CORBA, a client first looks up the object (server) it wants to communicate with. The ORB, as a result of the lookup operation, returns an object reference (a handle) of the server to the client. The client then uses the object reference to invoke operations on the object as a function call in the chosen programming language. The ORB intercepts the client request, collects the information about the operation and the request parameter values, encodes it in IIOP, and sends it to the object (server). The ORB on the object side (server) translates the request into a programming language specific function call on the server object. The server object then processes the request and returns a response, if any. The ORB intercepts the response, encodes the response and its parameters into IIOP, and sends it to the client. The ORB on the client side then returns the response to the client as the return value of the function call originally made as part of issuing the request.
In many cases, CORBA gateways to specific services may be developed to manage specific network traffic, such as requests and events. Typically, these gateways are designed as single-threaded programs. However, increasingly, the benefits of multi-threading are desired for applications and server programs related to network management.
Multi-Threaded Applications
A thread is an encapsulation of the flow of control in a program. Most programs are single-threaded; they only execute one path through their code at a time. Multi-threaded programs may have several threads running through different code paths simultaneously. Multi-threading typically means sharing a single CPU between multiple threads in a way designed to minimize the time required to switch threads. This is accomplished by sharing as much as possible of the program execution environment between the different threads so that very little state needs to be saved and restored when changing threads. Furthermore, if a computer has multiple CPUs and a program has multiple threads, multiple threads of the program may be run on separate CPUs concurrently. Thus multi-threading allows applications to scale with the number of CPUs, as well.
Multi-threading differs from multi-tasking in that threads share more of their environment with each other than do tasks under multi-tasking. Threads may be distinguished only by the value of their program counters and stack pointers while sharing a single address space and set of global variables. Often, there is very little protection of one thread from another, in contrast to multi-tasking. Multi-threading can thus be used for very fine-grained multi-tasking at the level of a few instructions and therefore can hide latency by keeping the processor busy after one thread issues a long-latency instruction on which subsequent instructions in that thread depend.
In a typical process in which multiple threads are used, zero or more threads may actually be running at any one time. This depends on the number of CPUs used by the computer on which the process is running, and also on how the threads system is implemented. A machine with n CPUs may run no more than n threads in parallel, but it may give the appearance of running many more than n threads simultaneously by sharing the CPUs among threads.
A context switch between two threads in a single process is considerably more efficient than a context switch between two processes. In addition, the fact that all data except for stack and registers are shared between threads makes them a natural vehicle for expressing tasks that may be broken down into subtasks that may be run cooperatively. Global variables and resources are shared between threads within the same process. Each thread has its own stack.
In many ways, the use of threads provides benefits over the use of processes in that threads are more efficient to create, switching between threads in the same process is much more efficient than switching processes, and there is easier sharing of resources between threads. Context switching among threads is very efficient in that there are no page or segment table manipulations, no flushing of the associative memory cache (when switching among threads sharing an address space), and no copying of data when exchanging messages among threads of the same address space.
As used herein, “thread-safe” refers to the property that a program may safely use or be used by multiple threads to avoid, for example, data inconsistencies, access collisions, coherency problems, and other errors. When multiple threads share resources, access to the resources should be synchronized to ensure thread-safety. One way this may be accomplished is through the use of a mutual exclusion (mutex) object. A mutual exclusion object allows multiple threads to synchronize access to shared resources. A mutex has two states: locked and unlocked. Once a mutex has been locked by a thread, other threads attempting to lock it will be blocked. When the locking thread unlocks (releases) the mutex, one of the blocked threads may acquire (lock) it and proceed. When managed object events and responses are delivered to a client manager application using multiple threads, synchronization and serialization of the event and response deliveries may become problematic in that the use of different threads to deliver sequential events may introduce chronological inconsistencies due to differing thread execution times. In other words, if a first event is sent using a first thread, and a subsequent second event is sent using a second thread, then depending upon the execution times of the two threads, the second event may actually be delivered prior to the first event.
The existing standards for TMN event management do not address the issue of access control to TMN events in a platform-independent interface. Currently, that is, mechanisms do not exist to readily manage which TMN manager applications have access to particular managed object events. In addition, clients are required to receive events only indirectly, by creating Common Management Information Service (CMIS) filters. These shortcomings may result in substantial loss of scalability and performance with regard to TMN object event management. Furthermore, the issues of client authentication, object-level access control, and security/audit trails for TMN system usage are not addressed by existing standards. Finally, the existing standards do not allow the clients the option to choose the format in which to receive events.
Therefore, improved systems and methods for managing objects over a network are desired.